Study of short-term feeding behaviour (STFB) could improve the understanding of variation in daily intake in dairy cows. STFB is generally measured in short bouts (e.g. visits to feeders) that are clustered in larger bouts (or meals). The value of bout analysis depends strongly on the choice of an appropriate bout. Before bouts can be grouped into meals, a meal criterion (MC, that is: the longest non-feeding interval accepted as part of a meal) must be estimated. Tolkamp and Kyriazakis (1999) criticised existing methods and recently developed a new technique to estimate meal criteria. These log-normal models were developed on basis of the idea that eating bouts end when animals are satiated (i.e., in a state of low feeding motivation) (Tolkamp and Kyriazakis, 1999). This implies that feed consumption during the relevant eating bout will result in a gradual increase in satiety. This will be associated with an increase in the probability of cows ending a bout. In this study we will analyse whether meals are a more biologically relevant unit of STFB than the short feeding bouts (i.e. visits) that are routinely recorded.